Probes your catalog to find search problems before shoppers do. It searches each sampled product by its own SKU, by brand, by a keyword from its name, and by a deliberate typo. No analytics needed.
Searches shoppers ran through the gateway. Your in-dashboard test searches are not counted. Each figure is compared with the preceding period of the same length, which is why the window stops at 45 days: there has to be a period before it to compare against. Figures are for the selected storefront only.
Counts queries and category pages together.
Gateway time for the Typesense search itself, without network.
Which storefront language shoppers searched in.
How many products came back, counted per search rather than per query.
Queries shoppers ran through the storefront widget / gateway (your in-dashboard test searches aren't counted). They tell you exactly what to curate. Counts are for the selected storefront only.
| query | lang | searches |
|---|
| query | lang | searches | zeros |
|---|
Two things the lists above cannot show. A search that returned three products failed the shopper about as badly as one that returned none, and it is the better lead, because the vocabulary is nearly right. Movement tells you what shoppers started or stopped looking for. Both follow the period picked in Search analytics.
| query | lang | searches | hits |
|---|
| query | lang | now | before | change |
|---|
Recorded from the storefront search and category pages. A facet nobody opens is clutter in the sidebar, and one that dominates belongs at the top of it. This starts collecting when the release that added it goes live, so it stays empty for a while even where the query lists above are full.
default means the shopper never touched the dropdown.
How far shoppers page. A lot of page 2 and beyond means the first page is not answering.
plentyONE referrer id the shopper arrived through. Pairs with the price list mapped to it under Settings, Pricing.
A category page or an empty search box is a browse, not a query, so it never appears in the query lists. The hour-of-day chart counts both.
Which fields a shopper's query searches, each field's weight, and its typo tolerance. Applies to the storefront immediately, with no reindex. Higher weight means more important; typos 0 means exact match only. Property fields from your facet schema can be made searchable here without being a facet. A newly added one takes effect after the next reindex.
| field | weight | typos | prefix |
|---|
prefix = search-as-you-type for that field. Turn it off for long text (description) so partial words don't flood results.
Which products are imported into search. Applies on the next full reindex.
Base currency and plentyONE salesPriceId of the selected storefront. Each storefront can sell off its own price list. Extra currencies are shared by all of them, and each one adds a price_<code> field to every collection. Adding or removing a currency needs a full reindex. Assigning a currency to a language (below) is live.
plentyONE prices per referrer, so a shopper arriving via Idealo or Google Shopping sees a different price than a direct visitor. The shop renders those prices itself; map them here and the search filter, the price sort and the autocomplete price agree with it. An unmapped referrer keeps the base price list. Each mapped list adds a price_l<id> field to every collection, so a change needs a full reindex.
Languages of the selected storefront. The primary is fixed (operators change it under Admin configuration). Adding a language creates its collection on the next full reindex; its translation-label inputs then appear under Facets & schema. Removing a language is handled by Heista support.
Per-tenant operator settings: quotas a merchant can't raise, sync cadence, lifecycle, backend, and the plentyONE connection.
| type | status | started | duration | stats |
|---|
One row per mandant of this plentyONE system. Each has its own collections, search token, sync state, category names and popularity; the catalog config (credentials, facets, relevance, visibility, curation) is shared. A new storefront is empty until its first full reindex. edit changes the display name and the mandant; the price list and currency are edited under Settings › Pricing with that storefront selected.
| key | name | mandant | price list | currency | status | languages | last full |
|---|
The key becomes part of every collection name (catalog_<slug>_<key>) and cannot be changed later, so pick it like a slug. Language codes are rejected: they would collide with a language collection. Languages are copied from the primary storefront; edit them under Settings with the storefront selected.
Public tokens the storefront widget or plugin presents to the gateway. Not a login. The token column shows the first and last characters, so you can match it against the value in your plugin config. Last used tells you which token is actually presenting itself. Leave origins empty to allow any origin while testing, and set them in production.
| label | storefront | token | origins | rate | created | last used |
|---|
Credentials a merchant uses to sign in to this portal (scoped to the tenant). Not for the storefront.
| label | active | last used |
|---|